Traditional voice telephony products are band-limited to 4 khz bandwidth with 8 kHz sampling, known as “narrowband”. These products include the telephone, data modems, and fax machines. Newer products aiming to achieve higher voice quality have doubled the sampling rate to 16 khz to encompass a larger 8 khz bandwidth, which is also known as “wideband” capable. The software implications of doubling the sampling rate are significant. Doubling the sampling rate not only requires doubling the processing cycles, but nearly doubling the memory used to store the data.
Doubling memory and processor cycles requirements is expensive because the memory and processing power footprints of DSPs are generally small. Implementing wideband support thus requires creativeness to optimize both memory and cycles.
Additionally, much of the software providing various functions and services, such as echo cancellation, dual-tone multi-frequency (DTMF) detection and generation, and call discrimination (between voice and facsimile transmission, for example), are written for only narrowband signals. Either software must be written for wideband signals, or the wideband signal down-sampled. Where the software is modified, the software should also be capable of integration with preexisting narrowband devices. Providing software for operation with both narrowband and wideband devices is complex and costly.
A scheme for down-sampling the wideband signal is presented in the co-pending application “Dual-Rate Single Band Communication System”.
Further limitations and disadvantages of conventional and traditional approaches will become apparent to one of skill in the art, through comparison of such systems with aspects of the present invention as set forth in the remainder of the present application with reference to the drawings.